1 import os, stat, time, struct, tempfile
2 from bup.helpers import *
6 INDEX_HDR = 'BUPI\0\0\0\2'
7 INDEX_SIG = '!IIIIIQII20sHII'
8 ENTLEN = struct.calcsize(INDEX_SIG)
10 FOOTLEN = struct.calcsize(FOOTER_SIG)
12 IX_EXISTS = 0x8000 # file exists on filesystem
13 IX_HASHVALID = 0x4000 # the stored sha1 matches the filesystem
14 IX_SHAMISSING = 0x2000 # the stored sha1 object doesn't seem to exist
16 class Error(Exception):
21 def __init__(self, ename, parent):
28 (ofs,n) = (f.tell(), len(self.list))
30 count = len(self.list)
31 #log('popping %r with %d entries\n'
32 # % (''.join(self.ename), count))
36 self.parent.count += count + self.count
40 def _golevel(level, f, ename, newentry):
41 # close nodes back up the tree
43 while ename[:len(level.ename)] != level.ename:
44 n = BlankNewEntry(level.ename[-1])
46 (n.children_ofs,n.children_n) = level.write(f)
47 level.parent.list.append(n)
50 # create nodes down the tree
51 while len(level.ename) < len(ename):
52 level = Level(ename[:len(level.ename)+1], level)
54 # are we in precisely the right place?
55 assert(ename == level.ename)
56 n = newentry or BlankNewEntry(ename and level.ename[-1] or None)
57 (n.children_ofs,n.children_n) = level.write(f)
59 level.parent.list.append(n)
66 def __init__(self, basename, name):
67 self.basename = str(basename)
73 return ("(%s,0x%04x,%d,%d,%d,%d,%d,%s/%s,0x%04x,0x%08x/%d)"
74 % (self.name, self.dev,
75 self.ctime, self.mtime, self.uid, self.gid,
76 self.size, oct(self.mode), oct(self.gitmode),
77 self.flags, self.children_ofs, self.children_n))
80 return struct.pack(INDEX_SIG,
81 self.dev, self.ctime, self.mtime,
82 self.uid, self.gid, self.size, self.mode,
83 self.gitmode, self.sha, self.flags,
84 self.children_ofs, self.children_n)
86 def from_stat(self, st, tstart):
87 old = (self.dev, self.ctime, self.mtime,
88 self.uid, self.gid, self.size, self.flags & IX_EXISTS)
89 new = (st.st_dev, int(st.st_ctime), int(st.st_mtime),
90 st.st_uid, st.st_gid, st.st_size, IX_EXISTS)
92 self.ctime = int(st.st_ctime)
93 self.mtime = int(st.st_mtime)
96 self.size = st.st_size
97 self.mode = st.st_mode
98 self.flags |= IX_EXISTS
99 if int(st.st_ctime) >= tstart or old != new \
100 or self.sha == EMPTY_SHA or not self.gitmode:
104 f = IX_HASHVALID|IX_EXISTS
105 return (self.flags & f) == f
107 def invalidate(self):
108 self.flags &= ~IX_HASHVALID
110 def validate(self, gitmode, sha):
113 self.gitmode = gitmode
115 self.flags |= IX_HASHVALID|IX_EXISTS
118 return not self.is_deleted()
120 def sha_missing(self):
121 return (self.flags & IX_SHAMISSING) or not (self.flags & IX_HASHVALID)
123 def is_deleted(self):
124 return (self.flags & IX_EXISTS) == 0
126 def set_deleted(self):
127 if self.flags & IX_EXISTS:
128 self.flags &= ~(IX_EXISTS | IX_HASHVALID)
131 return not self.is_fake()
134 return not self.ctime
137 return (cmp(a.name, b.name)
138 or -cmp(a.is_valid(), b.is_valid())
139 or -cmp(a.is_fake(), b.is_fake()))
142 f.write(self.basename + '\0' + self.packed())
145 class NewEntry(Entry):
146 def __init__(self, basename, name, dev, ctime, mtime, uid, gid,
147 size, mode, gitmode, sha, flags, children_ofs, children_n):
148 Entry.__init__(self, basename, name)
149 (self.dev, self.ctime, self.mtime, self.uid, self.gid,
150 self.size, self.mode, self.gitmode, self.sha,
151 self.flags, self.children_ofs, self.children_n
152 ) = (dev, int(ctime), int(mtime), uid, gid,
153 size, mode, gitmode, sha, flags, children_ofs, children_n)
156 class BlankNewEntry(NewEntry):
157 def __init__(self, basename):
158 NewEntry.__init__(self, basename, basename,
160 0, EMPTY_SHA, 0, 0, 0)
163 class ExistingEntry(Entry):
164 def __init__(self, parent, basename, name, m, ofs):
165 Entry.__init__(self, basename, name)
169 (self.dev, self.ctime, self.mtime, self.uid, self.gid,
170 self.size, self.mode, self.gitmode, self.sha,
171 self.flags, self.children_ofs, self.children_n
172 ) = struct.unpack(INDEX_SIG, str(buffer(m, ofs, ENTLEN)))
174 # effectively, we don't bother messing with IX_SHAMISSING if
175 # not IX_HASHVALID, since it's redundant, and repacking is more
176 # expensive than not repacking.
177 # This is implemented by having sha_missing() check IX_HASHVALID too.
178 def set_sha_missing(self, val):
180 oldval = self.sha_missing() and 1 or 0
182 flag = val and IX_SHAMISSING or 0
183 newflags = (self.flags & (~IX_SHAMISSING)) | flag
184 self.flags = newflags
187 def unset_sha_missing(self, flag):
188 if self.flags & IX_SHAMISSING:
189 self.flags &= ~IX_SHAMISSING
193 self._m[self._ofs:self._ofs+ENTLEN] = self.packed()
194 if self.parent and not self.is_valid():
195 self.parent.invalidate()
198 def iter(self, name=None, wantrecurse=None):
200 if dname and not dname.endswith('/'):
202 ofs = self.children_ofs
203 assert(ofs <= len(self._m))
204 assert(self.children_n < 1000000)
205 for i in xrange(self.children_n):
206 eon = self._m.find('\0', ofs)
210 basename = str(buffer(self._m, ofs, eon-ofs))
211 child = ExistingEntry(self, basename, self.name + basename,
214 or child.name.startswith(dname)
215 or child.name.endswith('/') and dname.startswith(child.name)):
216 if not wantrecurse or wantrecurse(child):
217 for e in child.iter(name=name, wantrecurse=wantrecurse):
219 if not name or child.name == name or child.name.startswith(dname):
221 ofs = eon + 1 + ENTLEN
228 def __init__(self, filename):
229 self.filename = filename
231 self.writable = False
235 f = open(filename, 'r+')
237 if e.errno == errno.ENOENT:
242 b = f.read(len(INDEX_HDR))
244 log('warning: %s: header: expected %r, got %r'
245 % (filename, INDEX_HDR, b))
247 st = os.fstat(f.fileno())
249 self.m = mmap_readwrite(f)
251 self.count = struct.unpack(FOOTER_SIG,
252 str(buffer(self.m, st.st_size-FOOTLEN, FOOTLEN)))[0]
258 return int(self.count)
260 def forward_iter(self):
262 while ofs+ENTLEN <= len(self.m)-FOOTLEN:
263 eon = self.m.find('\0', ofs)
267 basename = str(buffer(self.m, ofs, eon-ofs))
268 yield ExistingEntry(None, basename, basename, self.m, eon+1)
269 ofs = eon + 1 + ENTLEN
271 def iter(self, name=None, wantrecurse=None):
272 if len(self.m) > len(INDEX_HDR)+ENTLEN:
274 if dname and not dname.endswith('/'):
276 root = ExistingEntry(None, '/', '/',
277 self.m, len(self.m)-FOOTLEN-ENTLEN)
278 for sub in root.iter(name=name, wantrecurse=wantrecurse):
280 if not dname or dname == root.name:
290 if self.writable and self.m:
295 if self.writable and self.m:
298 self.writable = False
300 def filter(self, prefixes, wantrecurse=None):
301 for (rp, path) in reduce_paths(prefixes):
302 for e in self.iter(rp, wantrecurse=wantrecurse):
303 assert(e.name.startswith(rp))
304 name = path + e.name[len(rp):]
308 # FIXME: this function isn't very generic, because it splits the filename
309 # in an odd way and depends on a terminating '/' to indicate directories.
311 """Split a path into a list of elements of the file system hierarchy."""
313 l = [i+'/' for i in l[:-1]] + l[-1:]
315 l.pop() # extra blank caused by terminating '/'
320 def __init__(self, filename):
321 self.rootlevel = self.level = Level([], None)
326 self.filename = filename = realpath(filename)
327 (dir,name) = os.path.split(filename)
328 (ffd,self.tmpname) = tempfile.mkstemp('.tmp', filename, dir)
329 self.f = os.fdopen(ffd, 'wb', 65536)
330 self.f.write(INDEX_HDR)
340 os.unlink(self.tmpname)
344 self.level = _golevel(self.level, self.f, [], None)
345 self.count = self.rootlevel.count
348 self.f.write(struct.pack(FOOTER_SIG, self.count))
350 assert(self.level == None)
358 os.rename(self.tmpname, self.filename)
360 def _add(self, ename, entry):
361 if self.lastfile and self.lastfile <= ename:
362 raise Error('%r must come before %r'
363 % (''.join(e.name), ''.join(self.lastfile)))
364 self.lastfile = e.name
365 self.level = _golevel(self.level, self.f, ename, entry)
367 def add(self, name, st, hashgen = None):
368 endswith = name.endswith('/')
369 ename = pathsplit(name)
371 #log('add: %r %r\n' % (basename, name))
375 (gitmode, sha) = hashgen(name)
376 flags |= IX_HASHVALID
378 (gitmode, sha) = (0, EMPTY_SHA)
380 isdir = stat.S_ISDIR(st.st_mode)
381 assert(isdir == endswith)
382 e = NewEntry(basename, name, st.st_dev, int(st.st_ctime),
383 int(st.st_mtime), st.st_uid, st.st_gid,
384 st.st_size, st.st_mode, gitmode, sha, flags,
388 e = BlankNewEntry(basename)
394 def add_ixentry(self, e):
395 e.children_ofs = e.children_n = 0
396 self._add(pathsplit(e.name), e)
398 def new_reader(self):
400 return Reader(self.tmpname)
403 def reduce_paths(paths):
408 if stat.S_ISDIR(st.st_mode):
411 xpaths.append((rp, p))
416 for (rp, p) in xpaths:
417 if prev and (prev == rp
418 or (prev.endswith('/') and rp.startswith(prev))):
419 continue # already superceded by previous path
420 paths.append((rp, p))
422 paths.sort(reverse=True)
427 def __init__(self, iters):
431 # FIXME: doesn't remove duplicated entries between iters.
432 # That only happens for parent directories, but will mean the
433 # actual iteration returns fewer entries than this function counts.
434 return sum(len(it) for it in self.iters)
438 l = [iter(it) for it in self.iters]
439 l = [(next(it),it) for it in l]
440 l = filter(lambda x: x[0], l)
444 if not (count % 1024):
445 progress('bup: merging indexes (%d/%d)\r' % (count, total))
450 if e.name != lastname:
457 log('bup: merging indexes (%d/%d), done.\n' % (count, total))